Dismiss
There will be a system update on Friday, May 5th, 6 PM ET. You may experience a brief service interruption.
Browse Prior Art Database

RING BUFFER NFS FOR UNIX HOSTS

IP.com Disclosure Number: IPCOM000008814D
Original Publication Date: 1998-Sep-01
Included in the Prior Art Database: 2002-Jul-16
Document File: 2 page(s) / 101K

Publishing Venue

Motorola

Related People

Mandeep Dhami: AUTHOR

Abstract

This paper describes ring buffer based NFS that solves many annoying problems with creating stable integrated systems on traditional UNIX hosts. deleting any older backup, and hence creating space), and create a new file. Luckily this works for most UNIX logfiles, but it suffers as the oper- ations it uses are not atomic and the size is "time" and not "bytes" (hence a large number of writes in a short time cannot be handled grace- fully) and it only works if all applications are "well behaved" (i.e open by name before every write).

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 50% of the total text.

Page 1 of 2

0 M

MOTOROLA Technical Developments

-

RING BUFFER NFS FOR UNIX HOSTS

by Mandeep Dhami

  This paper describes ring buffer based NFS that solves many annoying problems with creating stable integrated systems on traditional UNIX hosts.

deleting any older backup, and hence creating space), and create a new file. Luckily this works for most UNIX logfiles, but it suffers as the oper- ations it uses are not atomic and the size is "time" and not "bytes" (hence a large number of writes in a short time cannot be handled grace- fully) and it only works if all applications are "well behaved" (i.e open by name before every write).

  For many third party applications, the last require- ment is more of an exception rather than the rule. If that is not the case, opening process has a handle to a file which might have been renamed later or even deleted. Only to clear that is by overwriting and hop- ing no data is lost. Messy at the very least!

4. Final solution is, of course, to rewrite the applica- tion to use the disk in a ring buffer pattern. Unfortunately, using standard UNIX file seman- tics this is not as simple as it sounds. And also as this is now application specific code, it needs to be linked into every application that needs it (and re-linked when a bug is fixed etc.).

1 .O INTRODUCTION

  Traditional UNIX hosts need a reasonable amount of periodic system maintenance to be available for extended periods of time. The most common problem to bring down a well engineered system is the operator forgetting to archive (and delete) some UNIX/third- party logfile which then fills up the disk leading to many related problems. This problem is further wors- ened by integrating multiple third party tools that were not designed with the same amount of care for resource usage.

  Many ad-hoc solutions exist to fix specific prob- lems or applications. Yet when one attempts to inte- grate new third party tools, one invariably finds some logtile that grows regularly with no guaranteed method which will assure that files can be archived without loss of important data.

2.0 TRADITIONAL SOLUTIONS

Traditional solutions to this problem are:

1. Use disk partitioning to isolate "critical applica- tions" from other "illbehaved" ones and hence limit the size of failure to a s...